MyAIProjects

🔥 1) Argumned BI: Tableau Pulse

Role: Data Product Owner
**Technologies: Tableau Pulse - Tableau Cloud - AWS - ** AI models & models: LLMs · Machine Learning · NPL

• Scope: Sales data mart
• Status: deployed in Prod and scaled to 120 business users

🧩 Overview

Implementing GenAI solutions focused on data analytics to enhance insights, accelerate development, and drive innovation across sales reporting. Led Proofs of Concept with major GenAI vendors in the data visualization environment to test business usability and scalability, contributing to the definition of the AI strategy for sales capabilities.

Scope: Sales data mart Status: deployed in prod and scaled to 50 business users

📎 Project Files

• 📊 Dashboard mock-ups

📊 Key Outcomes

• The adoption is higher know-how within the Data & Analytics Team, and with knowledge of the database
• Low flexibility in the interaction between user and data: Robust but Constrained: The platform’s robustness restricts flexibility

🔥 2) Scoring KPI for FAQ article (Customer and Satisfaction department)

Role: Product Manager / Data Lead
Technologies: Python - Dataiku - Tableau ** AI models: FAISS · LLM · Machine Learning · Similarity Scoring · Clustering
AI Hackathon – Team: B. Pinuela, J. Sabatés, O. Díaz, G. Lucca, I. Yurchurk, E. Llorens

🧩 Overview

As part of the Schneider Electric “Knowledge Cup – AI Hackathon”, our team developed an AI-driven solution to digitize and structure the company’s FAQ knowledge base, which was highly unstructured and difficult to maintain. The goal was to enable smarter search, automation opportunities, and improved knowledge management across product lines.

🚀 My Contributions

• Defined the product scope and led the functional design for two use cases:
  • UC1 – FAQ Similarity Scoring at title level
  • UC2 – FAQ-to-Document Semantic Matching
• Structured unorganized HTML-based FAQ content into consistent fields (Product Line, Environment, Issue, Resolution).
• Designed the end-to-end pipeline: data preparation → validation → LLM scoring → FAISS similarity → reporting.
• Coordinated and built dashboard mock-ups for similarity interpretation and clustering visualization.
• Ensured alignment between technical development and business expectations.

📊 Key Outcomes

• Developed a semantic similarity scoring engine (title-to-title, answer-to-answer, FAQ-to-document).
• Enabled clustering of related FAQs for faster expert review.
• Included automatic detection of broken links and inconsistent metadata.
• Avoided word-to-word comparison issues by introducing LLM-based contextual scoring.
• Prepared a scalable baseline to extend the solution across languages and additional document types.

📈 Proposed Next Steps

• Integrate a translation module (FR ↔ EN).
• Standardize FAQ structure globally (Product Line / Issue / Resolution as required fields).
• Extend clustering capabilities by product context and environment.
• Apply the solution to other document families with similar metadata formats.

📎 Project Files

• 📄 Hackathon Presentation (PDF/PPT) → *Add link (WIP) *

• 📊 Dashboard mock-ups → * screenshots*

• WIP

.............................................................

🔥 3) Argumned BI: Thoughtspot (PoC) - Argentic AI and conversational BI for Data & Analytics

Role: Data Product Owner
Technologies: AWS - Thoughtspot - Spotter Agent - Spotter Classic ** AI tech & models:LLMs · Machine Learning · NPL · RAG

• Scope: Sales data mart
• Status: deployed in Prod for testing purposes

🧩 Overview

Led Proofs of Concept with major GenAI vendors in the data visualization environment to test business usability and scalability, contributing to the definition of the AI strategy for sales capabilities. Test GenAI solutions focused on data analytics to enhance insights, accelerate development, and drive innovation across the sales reporting scope.

📎 Project Files

• 📊 Dashboard mock-ups (dummy data)

📊 Key Outcomes

• High flexibility in the interaction between user and data thanks to the Spotter Agent
• Powerful NPL model with a good semantic layer that leverages the analytical capability of the user ................................................................

🔥 4) ML: Recommendation System for Installed base product (MRO)

Role: Business Analyst
Technologies: Python - AWS - SQL ** AI tech & models: Machine Learning

• Scope: AI CoE HUB / BU: Power system/field services
• Data source: Salesforce (CASE) - SAP (work orders)
• Status: Deployed

🧩 Overview

Develop an AI‑driven system that can predict and recommend the maintenance actions or replacement cycles of installed base machines or spare parts. The objective was to provide the field sales team with a tool that could tell them when to contact a client to offer a new machine.

🚀 My Contributions: Case model created / Work Order relation

Problem framing: “Operators lack visibility into upcoming maintenance needs and upgrade opportunities across the installed base.” “Decision-making is reactive, leading to downtime and lost revenue.”

Recommendation engine combines:ML predictions - Business rules - Safety constraints - Regulatory requirements

..................................................

🔥 5) ML: Recommendation System - Web Activity Digest MVP

Role: Product Manager Technologies: Phyton - AWS ** AI tech & models: Machine Learning

• Scope: AI CoE HUB
• Status: Deployed

🧩 Overview

Develop a model recommendation system engine (model) that informs an account manager about products that a specific contact/client is interested in.

Stream 1 – Web Activity Digest MVP

🔥 6): Machine Learning model- recommendation system for Sales Team

1. Scoping and framing the problem statement: use case elaboration
2. Data Research and exploration: exploring the data available on the database level and data sources
3. IT architecture for scalable machine learning models: What is the right IT architecture to scale models

Product Evolution